A congestion-aware medium access control protocol for multi-rate ad-hoc networks

Research output: Chapter in Book/Report/Conference proceedingsChapterpeer-review

Abstract

This paper investigates the problem of how to improve TCP performance in multi-rate Ad-hoc networks with congested links. To improve network performance, different rate adaptation algorithms, such as Automatic Rate Fallback (ARF) and Receiver-Based AutoRate (RBAR), have been proposed to adapt the data rate according to the current channel quality. Opportunistic Auto Rate (OAR) protocol is an optimisation for any existing rate adaptation algorithm which leads to a significant performance gain by providing temporal fairness. We analyze the reasons for the high performance gain obtained using OAR, and show that the OAR protocol does not work well for TCP communications in Ad-hoc networks where nodes use different data rates to communicate with each other (heterogeneous). Based on these analysis, we propose a Congestion Reactive Opportunistic Auto Rate (CROAR) protocol, which is a new rate adaption enhancement tailored to improve TCP performance in heterogeneous multi-hop Ad-hoc networks. Extensive simulations show that CROAR, compared to OAR and RBAR, produces significant throughput and end-to-end transmission latency improvements while only marginally relaxing temporal fairness.

Original languageEnglish
Title of host publicationProceedings - The 31st IEEE Conference on Local Computer Networks, LCN 2006
Pages97-104
Number of pages8
DOIs
Publication statusPublished - 2006
Event31st Annual IEEE Conference on Local Computer Networks, LCN 2006 - Tampa, FL, United States
Duration: 14 Nov 200616 Nov 2006

Publication series

NameProceedings - Conference on Local Computer Networks, LCN

Conference

Conference31st Annual IEEE Conference on Local Computer Networks, LCN 2006
Country/TerritoryUnited States
CityTampa, FL
Period14/11/0616/11/06

Fingerprint

Dive into the research topics of 'A congestion-aware medium access control protocol for multi-rate ad-hoc networks'. Together they form a unique fingerprint.

Cite this